隨著高科技產業快速發展,可靠的電力供應相當重要,若發生嚴重電力事故其損失龐大,故在台電供電無法絕對穩定的情況下,大型用戶多會自備發電機組,以提升電力可靠度。 本文即在探討同時考慮台電電費支出以及發電機發電成本情況下,如何制訂最佳契約容量值,以達到節省用電成本支出的效果。台電的電費結構對應不同時段、不同季節皆有不同的計算方式,加入自備發電機運轉之後,發電機的發電能力限制,以及功率因數或契約容量調整費的考量,將使訂定最佳契約容量值變得比較困難。為解決此一問題,本文乃提出改良型田口方法作為求解最佳契約容量值以及自備發電機發電量的演算法,在近乎無限多種契約容量值的組合中,找出可使總用電成本最低之最佳組合。 為驗證所提方法之可行性,本文使用某光電廠實際用電數據資料,包括台電用電度數、發電機組容量以及欲制訂契約容量月份的負載需求預測,透過所提改良型田口最佳化方法,依台電電費結構計算總用電成本,決定最佳契約容量與自備發電機組發電量,並和現行其他最佳化演算法比較。由模擬結果證實,該光電廠使用本文所提方法每個月平均可節省約12.95%的用電費用。
Following rapid development of high-technology industry, reliable power supply is quite important. As a severe power event happens, it may cause very huge losses. In the situation that power supply of the utilities is unable to be absolutely reliable, many customers thus have their self-owned generating units in order to improve the power reliability. The thesis studies how to set optimal capacity contract in consideration of expenditure of the utility power and cost of self-generating power at the same time. To be achieved is the savings of power expenses. The power tariff rate has different structures in different periods and seasons. With the self-owned generators included, it would be more difficult to set the optimal capacity contracts by observing simultaneously the output ranges of the self-owned generating units, power factors, and the fees of regulating capacity contracts. The thesis proposes improved Taguchi method to determine the best capacity contracts and dispatch the power output of the self-owned generating units. The objective is to search the solution which has the lowest electrical power expenses from almost infinite combinations. To verify the feasibility of the proposes method, the thesis employs the practical data from an optoelectronics factory, which includes the amounts of power consumption from the utilities, capacities of generating units, and load demand forecasting in the months of planning period. Then, searched in the proposed Improved Taguchi Method is the optimal solution with minimal total power expenses according to the tariff structures of Taipower. The results are then compared to the existing other algorithms. The simulation results reveal that about 12.95% of electrical power expenses per month in average can be saved through the proposed approach.